- Email: [email protected]
Do stock splits signal undervaluation?
Accepted Manuscript Do stock splits signal undervaluation? Mohammad A. Karim, Sayan Sarkar PII: DOI: Reference:
S2214-6350(16)00011-3 http://dx.doi.org/10.1016/j.jbef.2016.01.004 JBEF 64
To appear in:
Journal of Behavioral and Experimental Finance
Received date: 1 November 2015 Revised date: 4 December 2015 Accepted date: 3 January 2016 Please cite this article as: Karim, M.A., Sarkar, S., Do stock splits signal undervaluation?. Journal of Behavioral and Experimental Finance (2016), http://dx.doi.org/10.1016/j.jbef.2016.01.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
*Manuscript
Do stock splits signal undervaluation?
Mohammad A Karim* Assistant Professor of Finance Division of Finance and Economics Marshall University College of Business CH 259 Huntington, WV 25755, USA Phone: 304-840-5894, Email: [email protected]
Sayan Sarkar The University of Texas at El Paso Economics and Finance Department College of Business Administration Rm. 247 El Paso, TX 79968-0543, USA Email: [email protected]
*Corresponding Author
Do stock splits signal undervaluation?
Abstract Signaling theory of stock splits postulates that managers use stock splits to convey favorable private information to the market about the fair value of the firm and thus attempts to reduce or eliminate undervaluation. In this paper, we investigate this proposition using three different misvaluation estimates. Contrary to the undervaluation hypothesis, we find that split firms are overvalued, rather than undervalued for the seven years surrounding split announcements. Moreover, the overvaluation reaches its peak in the split announcement year and declines in the post-split period. Overall, our results suggest that firms do not use stock splits to signal undervaluation.
Keywords: Stock Splits, Equity Mispricing
JEL classification: G14, G32
Page 2 of 16
1. Introduction A number of studies suggest that one potential explanation for stock splits is that they act as a signal to the market. If there is an information asymmetry between insiders and outside investors about the future performance, managers may use stock splits to convey favorable private information about the current value of the firm (Ikenberry et al., 1996, Brennan and Copeland, 1988b). As in any costly signaling equilibrium, for a separating equilibrium to exist, there must be a cost to those firms that attempt to mimic the signal. Stock splits may have direct and indirect costs. The direct costs may include investment banker fees, exchange listing fees, and the potential increase in trading costs (both in terms of the spread and to some extent, the commission). Generally, these costs are relatively small. The market reaction to the split announcement may be viewed as an indirect cost and this cost could prove to be significant.
The split announcement may only be an initial step in the signaling process. The announcement is likely to cause investors to examine the firm more closely (Grinblatt et al., 1984 refers to this concept as ‘attention hypothesis’ of stock splits). If the result of this examination reveals undervaluation, then the market reaction should be favorable. Consequently the market price of the stock should adjust toward ‘true’ or fundamental value and hence reduce or even eliminate the mispricing. The subsequent increase in stock price may easily offset the direct costs of split. However, if the market’s reaction is less favorable or negative, the split announcement becomes more costly.
It is well documented in the stock split literature that the market reacts positively to stock split announcements and, on average, split announcements earn 2-4% abnormal return in the few Page 3 of 16
days surrounding split announcements (e.g. Grinblatt et al., 1984; McNichols and Dravid, 1990; Ikenberry et al., 1996). This finding is consistent with the idea that stock splits signal undervaluation. On the other hand, long-term abnormal return and long-term operating performance of split firms is mixed. Some researchers find positive long-term (one to three year period) abnormal stock return after splits (e.g., Ikenberry et al., 1996; Desai and Jain, 1997; Ikenberry and Ramnath, 2002), others find no evidence of such return (e.g., Fama et al., 1969; Lakonishok and Lev, 1987; and Asquith et al., 1989). Long term operating performance studies also report contradictory findings (see Ikenberry and Ramnath, 2002; Huang et al., 2009). Contradictory evidence of long term stock and operating performance casts doubt on the idea that stock splits signal undervaluation of the current market price.
In our study, we use three direct measures of misvaluation surrounding the split announcement year to examine the conjecture that stock splits signal undervaluation. We estimate these three measures using Residual Income Model (RIM hereafter), Rhodes-Kropf et al., (2005, RKRV hereafter), and industry median adjusted market-to-book ratios. Using a U.S. stock split sample for the period 1974-2007, we find that split firms are overvalued rather than undervalued in the years surrounding split announcements. Moreover, overvaluation reaches its peak in the split announcement year and then gradually declines in the post announcement period. Our study contributes to the existing split literature in two ways. First, it investigates mispricing of split firms using direct measures of misvaluation by decomposing market-to-book ratio into misvaluation and growth components whereas prior studies use debatable proxies of misvaluation such as market-to-book ratio or some form of Tobin’s Q. Second, our study
Page 4 of 16
provides evidence of the role of mispricing surrounding cosmetic corporate events such as stock splits.
The rest of the paper is organized as follows: section 2 describes the mispricing estimation methods and the sample, section 3 reports our findings, section 4 discusses the rationale of the decision to split by overvalued firms and section 5 presents our conclusions.
2. Sample and methodology
2.1 Measures of mispricing We use three misvaluation measures: RIM based, RKRV based, and industry adjusted misvaluation measures. All of these methods uses market-to-book ratio to measure misvaluation. RIM and RKRV decompose market-to-book ratio into misvaluation and growth option components using the following equation.
MB = M/V * V/B
(1)
M, V, and B represent market value, intrinsic value, and book value of equity respectively. M/V represents misvaluation, and V/B represents the growth option component. The previous equation can be re-written in log form as follows:
m – b = (m – v) + (v – b)
(2)
Page 5 of 16
where m, and b are observable while v is estimated using RIM and RKRV. Positive (negative) values of (m-v) represents overvaluation (undervaluation) of the firm in that given year.
RIM estimates v using the following equation:
(3) where x and r represents earnings and cost of capital respectively. The second term of on the right hand side of the equation (3) represents the sum of present values of residual incomes1.
On the other hand RKRV extends equation (2) by decomposing misvaluation component into firm specific and industry specific misvaluation components using the following equation:
(4)
Where i, j, and t represent firm, industry, and fiscal year while vit and vjt represent fundamental value and long-run value components respectively. Here firm specific misvaluation,
represents the
represents the industry specific misvaluation, and
represents the growth options component. We add firm and industry specific misvaluation together to calculate our total misvaluation measure. RKRV estimates fundamental
1
For a detailed discussion on RIM see D’Mello and Shroff (2000).
Page 6 of 16
value and long-run value by relating market value of equity with book value of equity and other accounting variables as described in the following equation:
(5)
where NI+ is absolute value of net income, I is an indicator variable for negative net income observations, and LEV represents leverage. Fundamental value is estimated using time-series average conditional regression coefficients and long-run value is estimated using the industry average of time series regression coefficients2.
Our final measure of misvaluation is the industry median adjusted market-to-book ratio, calculated using the following equation:
Industry adjusted misvaluation = MBit – Median (MBjt)
(6)
where i, j, and t represent firm, industry and fiscal year respectively. We define industry using Fama-French 48 industry classification.
2.2 Sample Our initial split sample consists of only ordinary common stocks from the CRSP database from 1974 to 2014. We utilize Lin et al. (2009) criteria to identify our split sample. To be included in our sample the pre-split stock price has to be at least $10, split factor has to be 1 or higher, and the CRSP Factor to Adjust Price (CRSP mnemonic: FACPR) has to be equal to the 2
For a detailed discussion on RKRV (2005) model see Rhodes-Kropf, Robinson, and Viswanathan (2005).
Page 7 of 16
CRSP Factor to Adjust Shares Outstanding (CRSP: FACSHR). We only consider the ‘pure’ stock splits where there is no cash dividends or distribution events within a month range of the split announcement. We also require that split firms have seven years of misvaluation data surrounding a split announcement: three years preceding the split announcement, the year of the split announcement, and three years succeeding the split announcement. Finally we require that all three measures of misvaluation can be calculated for all seven years surrounding split announcements. Our final sample of stock splits consists of 2,065 stock splits from 1974 to 2007.We restrict our final split sample up to the year 2007 due to the following two reasons. First, to calculate the intrinsic value for a given year, RIM requires four succeeding years’ data. As a result we are able to calculate RIM based misvaluation measures up to 2010. Although RKRV and industry market-to-book based misvaluation measures are not subject to this restriction, for consistency we keep all of our misvaluation measures up to 2010. Second, we also require that misvaluation measures are available three years succeeding the split announcement year. This limits our split sample to 2007.
Table 1 Distribution of stock splits by year and industry. Panel A: Split distributions by fiscal year n Fiscal year % 1974 3 0.1 1975 7 0.3 1976 14 0.7 1977 20 1.0 1978 28 1.4 1979 34 1.6 1980 53 2.6 1981 79 3.8 1982 34 1.6 1983 89 4.3 1984 47 2.3 1985 67 3.2 1986 72 3.5 1987 92 4.5 Page 8 of 16
Fiscal year 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
n 45 65 53 62 95 106 87 108 112 116 76 75 54 86
% 2.2 3.1 2.6 3.0 4.6 5.1 4.2 5.2 5.4 5.6 3.7 3.6 2.6 4.2
1988 1989 1990
31 44 36
1.5 2.1 1.7
2005 2006 2007 Total
71 55 49 2,065
3.4 2.7 2.4 100.0
Panel B: Split distribution by Fama-French 12 industry classification Industry n % Consumer Non-Durables 105 5.1 Consumer Durables 29 1.4 Manufacturing 239 11.6 Energy 47 2.3 Chemicals and Allied Products 24 1.2 Business Equipment (Computers, software etc.) 608 29.4 Telephone and Television Transmission 28 1.4 Utilities 31 1.5 Wholesale, Retail and some services 298 14.4 Healthcare, Medical equipment and drugs 204 9.9 Finance 194 9.4 Others (Mines, Construction, Entertainment etc.) 258 12.5 Total 2,065 100.0 Note: Panel A reports the distribution of splits by fiscal year and panel B reports the distribution of splits based on Fama and French 12 industry classification from 1974 to 2007. All split information is from the CRSP database. To be included in the split sample pre-split stock price has to be at least $10, a stock must have a share code of 10 or 11 with split factor of 1 or higher, with the CRSP Factor to Adjust Price (FACPR) equal to the CRSP Factor to adjust Shares Outstanding (FACSHR).
Our final split sample consists of 1,152 unique firms with 2,065 splits. Table1 describes the yearly and industry distribution where industry is defined using Fama-French 12 industry classification3. Panel A indicates that the splits are well distributed across all years with the highest number of splits in the year 2000 (116 splits). Panel B shows that business equipment industry has the largest number of splits (608 splits) compared to any other industry.
3. Results
3
See http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_12_ind_port.html for details.
Page 9 of 16
First we examine the misvaluation measures in the seven years surrounding the split announcement year. In table 2 we report both mean and median misvaluation estimates for all the three methods.
Table 2: Mispricing measures surrounding split announcements for splitting firms Misvaluation measures Relative year RIM RKRV Industry adjusted Mean Median Mean Median Mean Median -3 0.01 0.05 * 0.14 ** 0.11 ** 1.43 ** 0.35 ** -2 0.03 * 0.11 ** 0.18 ** 0.15 ** 1.20 ** 0.44 ** -1 0.30 ** 0.34 ** 0.34 ** 0.30 ** 1.91 ** 0.87 ** 0 0.57 ** 0.57 ** 0.31 ** 0.26 ** 2.43 ** 1.24 ** +1 0.46 ** 0.49 ** 0.24 ** 0.21 ** 2.21 ** 0.76 ** +2 0.37 ** 0.41 ** 0.20 ** 0.16 ** 1.31 ** 0.53 ** +3 0.30 ** 0.34 ** 0.18 ** 0.16 ** 1.09 ** 0.37 ** Note: This table reports the measures of misvaluation for splitting firms seven years surrounding split announcements. Misvaluation measures are calculated using Residual Income Model, Rhodes-Kropf model and industry median adjusted mispricing model. Industry is defined using Fama French 48 industry classification for the median adjusted model. Non-split firms are non -splitting Compustat firms whose mispricing can be calculated using three mispricing methods. ** and * represents significance level at 1% and 5% respectively using t-test and Wilcoxon rank sum test.
We find that the mean and the median of all three misvaluation estimates are positive and significant in all seven years surrounding the split announcement year. This finding is contrary to the undervaluation hypothesis of stock splits and suggests that split firms are overvalued rather than undervalued surrounding the split announcement year. Moreover, if we examine the pattern of the mispricing we see that overvaluation exists three years preceding the announcement year, gradually increases up to the announcement year (or the preceding year of the announcement year in case of RKRV), then subsequently declines after the split announcement. According to RIM and industry adjusted misvaluation measures, the largest overvaluation is in the split announcement year (t=0). RKRV finds that the year preceding the split announcement year (t = 1) is the most overvalued year, and that the split announcement year itself (t=0) has the second largest overvaluation measure compared with any other years surrounding the announcement. Page 10 of 16
Overall these findings suggest that splitting firms are overvalued rather than undervalued surrounding split announcement years and firms decide to split during the period when overvaluation reaches its peak.
There is a possibility that overvaluation is a market wide phenomenon. It is not unique to firms that splits, non-splitting firms are also overvalued during periods of significant number of splits. In order to examine this, we consider a non-splitting sample of all Compustat firms for the period of 1974 to 2007. We use similar filters as splitting firms to select our non-splitting sample firms except that these firms do not split during our sample period. We consider only those non- splitting firms where we can estimate all three misvaluation measures. We find a sample of 67,307 firm-year observations of non-splitters for that period.
Table 3: Percentage of mispriced split and non-split firms surrounding split announcements % of mispriced firms RIM RKRV Industry adjusted Split Non-split Split Non-split Split Non-split Relative year firms firms firms firms firms firms -3 53.4 46.9 58.8 48.9 63.0 45.7 -2 56.8 48.1 63.3 50.2 65.0 45.7 -1 68.8 49.7 75.2 52.8 75.9 45.8 0 81.9 50.7 72.9 53.6 80.4 45.8 +1 76.4 50.3 67.6 54.1 73.8 45.8 +2 72.1 50.4 64.8 54.8 66.2 46.0 +3 69.6 49.6 64.1 55.5 62.3 46.1 Note: This table reports the percentage of overvalued split and non-split firms in seven years surrounding split announcement year. Overvaluation measures are calculated using Residual Income Model, RhodesKropf model and industry median adjusted mispricing model. Industry is defined using Fama French 48 industry classification for the median adjusted model. Non-split firms are non -splitting Compustat firms whose mispricing can be calculated using three mispricing methods.
We then proceed to identify overvalued (both splitting and non-splitting) firms using the following procedure : for any given firm-year, if a firm has positive misvaluation estimate then we consider that firm as overvalued in that year. We then calculate the percentage of overvalued splitting and non-splitting firm-year corresponding to all seven years surrounding split Page 11 of 16
announcements. Table 3 reports our findings. We find that a significantly larger percentage of splitting firms are overvalued compared to non-splitting firms. Similar to the trend shown in table 2, we find that the percentage of overvalued splitting firms increases gradually up to the split announcement year using RIM and Industry adjusted misvaluation methods and decreases thereafter. In case of RKRV method, percentage of overvalued firms is the highest in pre-split year. However, no such trend exists for non-splitting firms. These findings suggest that marketwide overvaluation is not responsible for splitting firms’ overvaluation.
To examine whether overvaluation results hold in a multivariate setting we conduct logistic regression analysis on the determinants of stock split. The dependent variable is a binary variable coded 1 if the firm splits and 0 for non-splitting firms. We use announcement year market-tobook decomposition components including mispricing measures as our explanatory variables. We use several control variables, documented in earlier literature, that affect split decision. Pre1997 and post2000 are dummy variables for minimum tick changes, traderange is a dummy variable that takes a value of 1 if the actual share price is 50% greater than the predicted price (estimated following Dyl and Elliott, 2006) and 0 otherwise. Stock appreciation is the ratio of the t-1 year-end share price over the t-3 year-end share price to capture the price appreciation in presplit period, and Shareholders represents number of common shareholders. Our results are reported in table 4.
Across all three methods, our split announcement year misvaluation measures are positive and significant. RIM and RKRV based misvaluation measures are significant at the 1 % level and industry adjusted measure is significant at 10% level (p value of 0.0816). Our results
Page 12 of 16
indicate that firms that are overvalued, are more likely to split their shares. We also use overvaluation and growth option dummies as predictor instead of raw measures of market-tobook components for robustness. The overvaluation (growth) dummy takes a value of 1 if the raw misvaluation (growth option) measure is positive, and zero otherwise. Our results indicate that overvaluation in a split announcement year increases the likelihood of stock splits. Our findings are robust to many different sample specifications (5-year instead of seven-year window surrounding announcement, and cleaner sample of only one split announcement surrounding five or seven years of split announcement year).
Table 4 Logistic regressions on the determinants of stock split. Dependent variable RIM RKRV Split = 1, Non-split = 0 Intercept -3.407 *** Misvaluation 1.603 *** Growth 1.071 *** Overvaluation_Dummy Growth_Dummy
-3.605 ***
Log_AT pre1997 post2000 Traderange StockApp Shareholders
-0.076 *** -0.506 *** -0.719 *** -1.041 *** -0.002 -0.001 ***
0.028 -0.046 -0.625 *** -1.160 *** -0.002 -0.001 ***
-4.209 *** 0.852 *** 1.775 ***
1.815 *** 0.825 *** 0.113 *** -0.074 -0.781 *** -1.059 *** -0.002 -0.001 ***
Industry adjusted
-4.456 ***
-1.851 *** -2.624 *** 0.001
1.000 *** 2.416 ***
1.545 ***
-0.050 *** -0.626 *** -0.820 *** -1.006 *** -0.002 -0.001 ***
-0.023 * -0.569 *** -0.836 *** -0.959 *** 0.003 -0.001 ***
-0.052 *** -0.600 *** -0.813 *** -1.013 *** 0.002 -0.001 ***
-2 log L 10713 11606 10925 11843 12546 11780 Pseudo R2 0.066 0.042 0.061 0.036 0.017 0.038 Note: This table reports the logit regressions to analyze the split decision. The dependent variable is a binary variable coded 1 if the firm splits and 0 for non-splitting firms. Misvaluation and Growth represents mispricing and growth opportunities measure calculated by decomposing market-to-book ratio. Pre1997 and post2000 are control variables for minimum tick changes, traderange is a dummy variable that takes a value of 1 if the actual share price is 50% greater than the predicted price (estimated following Dyl and Elliott, 2006) and 0 otherwise, stock appreciation is the ratio of the t-1 year-end share price over the t-3 year-end share price to capture the price appreciation in pre-split period and Shareholders represents number of common shareholders. ***, **, * represents significance level at 0.1%, 1%, and 5% level, respectively.
Page 13 of 16
4. Why overvalued firms split the stocks We find that split firms are overvalued rather than undervalued, in pre-split years. Hence, splitting does not seem to be related to signaling undervaluation. So, why do overvalued firms split their stocks? Alternative theories attempting to explain splitting behavior, such as the liquidity and marketability hypotheses of stock splits, do not explicitly relate splitting behavior to overvaluation4. However, two other existing explanations for stock splits may have bearing on the fact that overvalued firms are more likely to split their stock. First, stock splits may be a response by overvalued firms (which tend to have higher stock prices) to keep their stock prices to a level that conforms to the customs or norms of the market. Weld et al., (2009) document that since the 1930s, equally weighted and value weighted average share prices of U.S. firms remain constant at about $25 and $35, respectively. Hence, firms may conform to this norm through stock splits. Second, stock splits may be the result of managerial opportunistic behavior. This behavioral explanation depends on the well documented fact that the market usually reacts positively to stock split announcement5.Two possible scenarios may illustrate this. In the first scenario, managers of overvalued firms may split their stocks not to reduce information asymmetry, but to serve the interests of the shareholders, when such actions may results in economic gain. Guo et al., (2008) finds evidence of such stock splits in a mergers and acquisition (M&A) setting. They document that acquiring firms use stock splits to manipulate their equity values prior to the acquisition announcements, especially when the method of payment is in stock and when the deals are large. Due to the positive price reaction after split announcement acquirer shareholders gain wealth at the expense of target shareholders. Another possibility is that, managers of
4 5
Weld et al. (2009) provides an excellent review of the literature on these (and other) explanations for stock splits. See Fama et al. (1969).
Page 14 of 16
overvalued firms split to serve their own interest, potentially at the expense of their shareholders. Devos et al., (2015) find that CEOs manipulate the timing of the stock splits in order to increase CEO compensation through executive stock and option grants. Typically stock options are granted with an exercise price equal to the closing price of the grant announcement day. Using a sample of stock splits where the CEOs stock option grants occur within twenty one days surrounding split announcement, they find that majority (80%) of these grants occur on or before the split announcement day. They also find significantly higher levels of selling (purchasing) activity immediately after (before) split announcement. Examining the role of overvaluation in firms split decision is beyond the scope of this paper. Future research may address this area.
5. Conclusion This study investigates the proposition of signaling hypothesis which states that stock split announcements convey favorable inside information and therefore signal undervaluation of the splitting firms. Using three different mispricing measures, our results indicate otherwise. Contradictory to the signaling hypothesis, our findings show that split firms are overvalued in pre-announcement years and that the overvaluation reaches its peak in the split announcement year as compared to the years surrounding the split announcements. Moreover, overvaluation gradually declines in post-split announcement years across all three methods. Our results support the idea that stock splits do not signal undervaluation to the market.
Acknowledgments We thank Erik Devos and Rishav Bista for helpful comments.
Page 15 of 16
References Asquith, P., Healy, P. and Palepu, K. (1989), “Earnings and Stock Splits”, The Accounting Review, Vol. 64 No. 3, pp. 387-403. Brennan, M. and Copeland, T. (1988b), “Stock splits, stock prices and transaction costs”, Journal of Financial Economics, Vol. 22, No. 1, pp. 83-101. Desai, H. and Jain, P. (1997), “Long-Run Common Stock Returns Following Stock Splits and Reverse Splits”, Journal of Business, Vol. 70, No. 3, pp. 409-433. Devos, E., Elliott, W. B. and Warr, R. S. (2015), “CEO opportunism?: Option grants and stock trades around stock splits”, Journal of Accounting and Economics, Vol. 60 No. 1, pp. 18-35. D’Mello, R., and Shroff, P. (2000) ‘Equity valuation and decisions related to repurchase tender offers: an empirical investigation’, The Journal of Finance, Vol. 55 No. 5, pp. 2399-2424. Dyl, E. and Elliott, W. (2006), “The share price puzzle”, The Journal of Business, Vol. 79 No. 4, pp. 2045-2066. Fama, E., Fisher, L., Jensen, M. and Roll, R. (1969), “The adjustment of stock prices to new information”, International Economic Review, Vol. 10 No. 1, pp. 1-21. Grinblatt, M., Masulis, R. and Titman, S. (1984), “The Valuation Effects of Stock Splits and Stock Dividends”, Journal of Financial Economics, Vol. 13 No. 4, pp.461-490. Guo, S., Liu, M. H. and Song, W. (2008), “Stock Splits as a Manipulation tool: Evidence from Mergers and Acquisitions”, Financial Management, Vol. 37 No. 4, pp. 695-712. Huang, G., Liano, K., Manakyan, H. and Pan, M. (2008), “The Information Content of Multiple Stock Splits”, Financial Review, Vol. 3, No. 4, pp. 543-567. Ikenberry, D., Rankine, G. and Stice, E. (1996), “What do stock splits really signal?”, Journal of Financial and Quantitative Analysis, Vol. 31, No. 3, pp. 357-375. Ikenberry, D. and Ramnath, S. (2002), “Underreaction to Self-Selected News Events: The Case of Stock Splits”, Review of Financial Studies, Vol. 15 No. 2, pp. 489-526. Lakonishok, J. and Lev B. (1987), “Stock splits and stock dividends: Why, who, and when”, The Journal of Finance, Vol. 42 No. 4, pp. 913-932. Lin, J., Singh, A. and Yu, W. (2009), “Stock splits, trading continuity, and the cost of equity capital”, Journal of Financial Economics, Vol. 93, No. 3, pp. 474-489. McNichols, M. and Dravid, A. (1990), “Stock dividends, stock splits, and signaling”, The Journal of Finance, Vol. 45 No. 3, pp. 857-879. Rhodes-Kropf, M., Robinson, D. and Viswanathan, S. (2005), “Valuation waves and merger activity: The empirical evidence”, Journal of Financial Economics, Vol. 77 No. 3, pp. 561–603. Weld, W., Michaely, R., Thaler, R. and Benartzi, S. (2009), “The nominal share price puzzle”, Journal of Economic Perspectives, Vol. 23 No. 2, pp. 121-142.
Page 16 of 16
S2214-6350(16)00011-3 http://dx.doi.org/10.1016/j.jbef.2016.01.004 JBEF 64
To appear in:
Journal of Behavioral and Experimental Finance
Received date: 1 November 2015 Revised date: 4 December 2015 Accepted date: 3 January 2016 Please cite this article as: Karim, M.A., Sarkar, S., Do stock splits signal undervaluation?. Journal of Behavioral and Experimental Finance (2016), http://dx.doi.org/10.1016/j.jbef.2016.01.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
*Manuscript
Do stock splits signal undervaluation?
Mohammad A Karim* Assistant Professor of Finance Division of Finance and Economics Marshall University College of Business CH 259 Huntington, WV 25755, USA Phone: 304-840-5894, Email: [email protected]
Sayan Sarkar The University of Texas at El Paso Economics and Finance Department College of Business Administration Rm. 247 El Paso, TX 79968-0543, USA Email: [email protected]
*Corresponding Author
Do stock splits signal undervaluation?
Abstract Signaling theory of stock splits postulates that managers use stock splits to convey favorable private information to the market about the fair value of the firm and thus attempts to reduce or eliminate undervaluation. In this paper, we investigate this proposition using three different misvaluation estimates. Contrary to the undervaluation hypothesis, we find that split firms are overvalued, rather than undervalued for the seven years surrounding split announcements. Moreover, the overvaluation reaches its peak in the split announcement year and declines in the post-split period. Overall, our results suggest that firms do not use stock splits to signal undervaluation.
Keywords: Stock Splits, Equity Mispricing
JEL classification: G14, G32
Page 2 of 16
1. Introduction A number of studies suggest that one potential explanation for stock splits is that they act as a signal to the market. If there is an information asymmetry between insiders and outside investors about the future performance, managers may use stock splits to convey favorable private information about the current value of the firm (Ikenberry et al., 1996, Brennan and Copeland, 1988b). As in any costly signaling equilibrium, for a separating equilibrium to exist, there must be a cost to those firms that attempt to mimic the signal. Stock splits may have direct and indirect costs. The direct costs may include investment banker fees, exchange listing fees, and the potential increase in trading costs (both in terms of the spread and to some extent, the commission). Generally, these costs are relatively small. The market reaction to the split announcement may be viewed as an indirect cost and this cost could prove to be significant.
The split announcement may only be an initial step in the signaling process. The announcement is likely to cause investors to examine the firm more closely (Grinblatt et al., 1984 refers to this concept as ‘attention hypothesis’ of stock splits). If the result of this examination reveals undervaluation, then the market reaction should be favorable. Consequently the market price of the stock should adjust toward ‘true’ or fundamental value and hence reduce or even eliminate the mispricing. The subsequent increase in stock price may easily offset the direct costs of split. However, if the market’s reaction is less favorable or negative, the split announcement becomes more costly.
It is well documented in the stock split literature that the market reacts positively to stock split announcements and, on average, split announcements earn 2-4% abnormal return in the few Page 3 of 16
days surrounding split announcements (e.g. Grinblatt et al., 1984; McNichols and Dravid, 1990; Ikenberry et al., 1996). This finding is consistent with the idea that stock splits signal undervaluation. On the other hand, long-term abnormal return and long-term operating performance of split firms is mixed. Some researchers find positive long-term (one to three year period) abnormal stock return after splits (e.g., Ikenberry et al., 1996; Desai and Jain, 1997; Ikenberry and Ramnath, 2002), others find no evidence of such return (e.g., Fama et al., 1969; Lakonishok and Lev, 1987; and Asquith et al., 1989). Long term operating performance studies also report contradictory findings (see Ikenberry and Ramnath, 2002; Huang et al., 2009). Contradictory evidence of long term stock and operating performance casts doubt on the idea that stock splits signal undervaluation of the current market price.
In our study, we use three direct measures of misvaluation surrounding the split announcement year to examine the conjecture that stock splits signal undervaluation. We estimate these three measures using Residual Income Model (RIM hereafter), Rhodes-Kropf et al., (2005, RKRV hereafter), and industry median adjusted market-to-book ratios. Using a U.S. stock split sample for the period 1974-2007, we find that split firms are overvalued rather than undervalued in the years surrounding split announcements. Moreover, overvaluation reaches its peak in the split announcement year and then gradually declines in the post announcement period. Our study contributes to the existing split literature in two ways. First, it investigates mispricing of split firms using direct measures of misvaluation by decomposing market-to-book ratio into misvaluation and growth components whereas prior studies use debatable proxies of misvaluation such as market-to-book ratio or some form of Tobin’s Q. Second, our study
Page 4 of 16
provides evidence of the role of mispricing surrounding cosmetic corporate events such as stock splits.
The rest of the paper is organized as follows: section 2 describes the mispricing estimation methods and the sample, section 3 reports our findings, section 4 discusses the rationale of the decision to split by overvalued firms and section 5 presents our conclusions.
2. Sample and methodology
2.1 Measures of mispricing We use three misvaluation measures: RIM based, RKRV based, and industry adjusted misvaluation measures. All of these methods uses market-to-book ratio to measure misvaluation. RIM and RKRV decompose market-to-book ratio into misvaluation and growth option components using the following equation.
MB = M/V * V/B
(1)
M, V, and B represent market value, intrinsic value, and book value of equity respectively. M/V represents misvaluation, and V/B represents the growth option component. The previous equation can be re-written in log form as follows:
m – b = (m – v) + (v – b)
(2)
Page 5 of 16
where m, and b are observable while v is estimated using RIM and RKRV. Positive (negative) values of (m-v) represents overvaluation (undervaluation) of the firm in that given year.
RIM estimates v using the following equation:
(3) where x and r represents earnings and cost of capital respectively. The second term of on the right hand side of the equation (3) represents the sum of present values of residual incomes1.
On the other hand RKRV extends equation (2) by decomposing misvaluation component into firm specific and industry specific misvaluation components using the following equation:
(4)
Where i, j, and t represent firm, industry, and fiscal year while vit and vjt represent fundamental value and long-run value components respectively. Here firm specific misvaluation,
represents the
represents the industry specific misvaluation, and
represents the growth options component. We add firm and industry specific misvaluation together to calculate our total misvaluation measure. RKRV estimates fundamental
1
For a detailed discussion on RIM see D’Mello and Shroff (2000).
Page 6 of 16
value and long-run value by relating market value of equity with book value of equity and other accounting variables as described in the following equation:
(5)
where NI+ is absolute value of net income, I is an indicator variable for negative net income observations, and LEV represents leverage. Fundamental value is estimated using time-series average conditional regression coefficients and long-run value is estimated using the industry average of time series regression coefficients2.
Our final measure of misvaluation is the industry median adjusted market-to-book ratio, calculated using the following equation:
Industry adjusted misvaluation = MBit – Median (MBjt)
(6)
where i, j, and t represent firm, industry and fiscal year respectively. We define industry using Fama-French 48 industry classification.
2.2 Sample Our initial split sample consists of only ordinary common stocks from the CRSP database from 1974 to 2014. We utilize Lin et al. (2009) criteria to identify our split sample. To be included in our sample the pre-split stock price has to be at least $10, split factor has to be 1 or higher, and the CRSP Factor to Adjust Price (CRSP mnemonic: FACPR) has to be equal to the 2
For a detailed discussion on RKRV (2005) model see Rhodes-Kropf, Robinson, and Viswanathan (2005).
Page 7 of 16
CRSP Factor to Adjust Shares Outstanding (CRSP: FACSHR). We only consider the ‘pure’ stock splits where there is no cash dividends or distribution events within a month range of the split announcement. We also require that split firms have seven years of misvaluation data surrounding a split announcement: three years preceding the split announcement, the year of the split announcement, and three years succeeding the split announcement. Finally we require that all three measures of misvaluation can be calculated for all seven years surrounding split announcements. Our final sample of stock splits consists of 2,065 stock splits from 1974 to 2007.We restrict our final split sample up to the year 2007 due to the following two reasons. First, to calculate the intrinsic value for a given year, RIM requires four succeeding years’ data. As a result we are able to calculate RIM based misvaluation measures up to 2010. Although RKRV and industry market-to-book based misvaluation measures are not subject to this restriction, for consistency we keep all of our misvaluation measures up to 2010. Second, we also require that misvaluation measures are available three years succeeding the split announcement year. This limits our split sample to 2007.
Table 1 Distribution of stock splits by year and industry. Panel A: Split distributions by fiscal year n Fiscal year % 1974 3 0.1 1975 7 0.3 1976 14 0.7 1977 20 1.0 1978 28 1.4 1979 34 1.6 1980 53 2.6 1981 79 3.8 1982 34 1.6 1983 89 4.3 1984 47 2.3 1985 67 3.2 1986 72 3.5 1987 92 4.5 Page 8 of 16
Fiscal year 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
n 45 65 53 62 95 106 87 108 112 116 76 75 54 86
% 2.2 3.1 2.6 3.0 4.6 5.1 4.2 5.2 5.4 5.6 3.7 3.6 2.6 4.2
1988 1989 1990
31 44 36
1.5 2.1 1.7
2005 2006 2007 Total
71 55 49 2,065
3.4 2.7 2.4 100.0
Panel B: Split distribution by Fama-French 12 industry classification Industry n % Consumer Non-Durables 105 5.1 Consumer Durables 29 1.4 Manufacturing 239 11.6 Energy 47 2.3 Chemicals and Allied Products 24 1.2 Business Equipment (Computers, software etc.) 608 29.4 Telephone and Television Transmission 28 1.4 Utilities 31 1.5 Wholesale, Retail and some services 298 14.4 Healthcare, Medical equipment and drugs 204 9.9 Finance 194 9.4 Others (Mines, Construction, Entertainment etc.) 258 12.5 Total 2,065 100.0 Note: Panel A reports the distribution of splits by fiscal year and panel B reports the distribution of splits based on Fama and French 12 industry classification from 1974 to 2007. All split information is from the CRSP database. To be included in the split sample pre-split stock price has to be at least $10, a stock must have a share code of 10 or 11 with split factor of 1 or higher, with the CRSP Factor to Adjust Price (FACPR) equal to the CRSP Factor to adjust Shares Outstanding (FACSHR).
Our final split sample consists of 1,152 unique firms with 2,065 splits. Table1 describes the yearly and industry distribution where industry is defined using Fama-French 12 industry classification3. Panel A indicates that the splits are well distributed across all years with the highest number of splits in the year 2000 (116 splits). Panel B shows that business equipment industry has the largest number of splits (608 splits) compared to any other industry.
3. Results
3
See http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_12_ind_port.html for details.
Page 9 of 16
First we examine the misvaluation measures in the seven years surrounding the split announcement year. In table 2 we report both mean and median misvaluation estimates for all the three methods.
Table 2: Mispricing measures surrounding split announcements for splitting firms Misvaluation measures Relative year RIM RKRV Industry adjusted Mean Median Mean Median Mean Median -3 0.01 0.05 * 0.14 ** 0.11 ** 1.43 ** 0.35 ** -2 0.03 * 0.11 ** 0.18 ** 0.15 ** 1.20 ** 0.44 ** -1 0.30 ** 0.34 ** 0.34 ** 0.30 ** 1.91 ** 0.87 ** 0 0.57 ** 0.57 ** 0.31 ** 0.26 ** 2.43 ** 1.24 ** +1 0.46 ** 0.49 ** 0.24 ** 0.21 ** 2.21 ** 0.76 ** +2 0.37 ** 0.41 ** 0.20 ** 0.16 ** 1.31 ** 0.53 ** +3 0.30 ** 0.34 ** 0.18 ** 0.16 ** 1.09 ** 0.37 ** Note: This table reports the measures of misvaluation for splitting firms seven years surrounding split announcements. Misvaluation measures are calculated using Residual Income Model, Rhodes-Kropf model and industry median adjusted mispricing model. Industry is defined using Fama French 48 industry classification for the median adjusted model. Non-split firms are non -splitting Compustat firms whose mispricing can be calculated using three mispricing methods. ** and * represents significance level at 1% and 5% respectively using t-test and Wilcoxon rank sum test.
We find that the mean and the median of all three misvaluation estimates are positive and significant in all seven years surrounding the split announcement year. This finding is contrary to the undervaluation hypothesis of stock splits and suggests that split firms are overvalued rather than undervalued surrounding the split announcement year. Moreover, if we examine the pattern of the mispricing we see that overvaluation exists three years preceding the announcement year, gradually increases up to the announcement year (or the preceding year of the announcement year in case of RKRV), then subsequently declines after the split announcement. According to RIM and industry adjusted misvaluation measures, the largest overvaluation is in the split announcement year (t=0). RKRV finds that the year preceding the split announcement year (t = 1) is the most overvalued year, and that the split announcement year itself (t=0) has the second largest overvaluation measure compared with any other years surrounding the announcement. Page 10 of 16
Overall these findings suggest that splitting firms are overvalued rather than undervalued surrounding split announcement years and firms decide to split during the period when overvaluation reaches its peak.
There is a possibility that overvaluation is a market wide phenomenon. It is not unique to firms that splits, non-splitting firms are also overvalued during periods of significant number of splits. In order to examine this, we consider a non-splitting sample of all Compustat firms for the period of 1974 to 2007. We use similar filters as splitting firms to select our non-splitting sample firms except that these firms do not split during our sample period. We consider only those non- splitting firms where we can estimate all three misvaluation measures. We find a sample of 67,307 firm-year observations of non-splitters for that period.
Table 3: Percentage of mispriced split and non-split firms surrounding split announcements % of mispriced firms RIM RKRV Industry adjusted Split Non-split Split Non-split Split Non-split Relative year firms firms firms firms firms firms -3 53.4 46.9 58.8 48.9 63.0 45.7 -2 56.8 48.1 63.3 50.2 65.0 45.7 -1 68.8 49.7 75.2 52.8 75.9 45.8 0 81.9 50.7 72.9 53.6 80.4 45.8 +1 76.4 50.3 67.6 54.1 73.8 45.8 +2 72.1 50.4 64.8 54.8 66.2 46.0 +3 69.6 49.6 64.1 55.5 62.3 46.1 Note: This table reports the percentage of overvalued split and non-split firms in seven years surrounding split announcement year. Overvaluation measures are calculated using Residual Income Model, RhodesKropf model and industry median adjusted mispricing model. Industry is defined using Fama French 48 industry classification for the median adjusted model. Non-split firms are non -splitting Compustat firms whose mispricing can be calculated using three mispricing methods.
We then proceed to identify overvalued (both splitting and non-splitting) firms using the following procedure : for any given firm-year, if a firm has positive misvaluation estimate then we consider that firm as overvalued in that year. We then calculate the percentage of overvalued splitting and non-splitting firm-year corresponding to all seven years surrounding split Page 11 of 16
announcements. Table 3 reports our findings. We find that a significantly larger percentage of splitting firms are overvalued compared to non-splitting firms. Similar to the trend shown in table 2, we find that the percentage of overvalued splitting firms increases gradually up to the split announcement year using RIM and Industry adjusted misvaluation methods and decreases thereafter. In case of RKRV method, percentage of overvalued firms is the highest in pre-split year. However, no such trend exists for non-splitting firms. These findings suggest that marketwide overvaluation is not responsible for splitting firms’ overvaluation.
To examine whether overvaluation results hold in a multivariate setting we conduct logistic regression analysis on the determinants of stock split. The dependent variable is a binary variable coded 1 if the firm splits and 0 for non-splitting firms. We use announcement year market-tobook decomposition components including mispricing measures as our explanatory variables. We use several control variables, documented in earlier literature, that affect split decision. Pre1997 and post2000 are dummy variables for minimum tick changes, traderange is a dummy variable that takes a value of 1 if the actual share price is 50% greater than the predicted price (estimated following Dyl and Elliott, 2006) and 0 otherwise. Stock appreciation is the ratio of the t-1 year-end share price over the t-3 year-end share price to capture the price appreciation in presplit period, and Shareholders represents number of common shareholders. Our results are reported in table 4.
Across all three methods, our split announcement year misvaluation measures are positive and significant. RIM and RKRV based misvaluation measures are significant at the 1 % level and industry adjusted measure is significant at 10% level (p value of 0.0816). Our results
Page 12 of 16
indicate that firms that are overvalued, are more likely to split their shares. We also use overvaluation and growth option dummies as predictor instead of raw measures of market-tobook components for robustness. The overvaluation (growth) dummy takes a value of 1 if the raw misvaluation (growth option) measure is positive, and zero otherwise. Our results indicate that overvaluation in a split announcement year increases the likelihood of stock splits. Our findings are robust to many different sample specifications (5-year instead of seven-year window surrounding announcement, and cleaner sample of only one split announcement surrounding five or seven years of split announcement year).
Table 4 Logistic regressions on the determinants of stock split. Dependent variable RIM RKRV Split = 1, Non-split = 0 Intercept -3.407 *** Misvaluation 1.603 *** Growth 1.071 *** Overvaluation_Dummy Growth_Dummy
-3.605 ***
Log_AT pre1997 post2000 Traderange StockApp Shareholders
-0.076 *** -0.506 *** -0.719 *** -1.041 *** -0.002 -0.001 ***
0.028 -0.046 -0.625 *** -1.160 *** -0.002 -0.001 ***
-4.209 *** 0.852 *** 1.775 ***
1.815 *** 0.825 *** 0.113 *** -0.074 -0.781 *** -1.059 *** -0.002 -0.001 ***
Industry adjusted
-4.456 ***
-1.851 *** -2.624 *** 0.001
1.000 *** 2.416 ***
1.545 ***
-0.050 *** -0.626 *** -0.820 *** -1.006 *** -0.002 -0.001 ***
-0.023 * -0.569 *** -0.836 *** -0.959 *** 0.003 -0.001 ***
-0.052 *** -0.600 *** -0.813 *** -1.013 *** 0.002 -0.001 ***
-2 log L 10713 11606 10925 11843 12546 11780 Pseudo R2 0.066 0.042 0.061 0.036 0.017 0.038 Note: This table reports the logit regressions to analyze the split decision. The dependent variable is a binary variable coded 1 if the firm splits and 0 for non-splitting firms. Misvaluation and Growth represents mispricing and growth opportunities measure calculated by decomposing market-to-book ratio. Pre1997 and post2000 are control variables for minimum tick changes, traderange is a dummy variable that takes a value of 1 if the actual share price is 50% greater than the predicted price (estimated following Dyl and Elliott, 2006) and 0 otherwise, stock appreciation is the ratio of the t-1 year-end share price over the t-3 year-end share price to capture the price appreciation in pre-split period and Shareholders represents number of common shareholders. ***, **, * represents significance level at 0.1%, 1%, and 5% level, respectively.
Page 13 of 16
4. Why overvalued firms split the stocks We find that split firms are overvalued rather than undervalued, in pre-split years. Hence, splitting does not seem to be related to signaling undervaluation. So, why do overvalued firms split their stocks? Alternative theories attempting to explain splitting behavior, such as the liquidity and marketability hypotheses of stock splits, do not explicitly relate splitting behavior to overvaluation4. However, two other existing explanations for stock splits may have bearing on the fact that overvalued firms are more likely to split their stock. First, stock splits may be a response by overvalued firms (which tend to have higher stock prices) to keep their stock prices to a level that conforms to the customs or norms of the market. Weld et al., (2009) document that since the 1930s, equally weighted and value weighted average share prices of U.S. firms remain constant at about $25 and $35, respectively. Hence, firms may conform to this norm through stock splits. Second, stock splits may be the result of managerial opportunistic behavior. This behavioral explanation depends on the well documented fact that the market usually reacts positively to stock split announcement5.Two possible scenarios may illustrate this. In the first scenario, managers of overvalued firms may split their stocks not to reduce information asymmetry, but to serve the interests of the shareholders, when such actions may results in economic gain. Guo et al., (2008) finds evidence of such stock splits in a mergers and acquisition (M&A) setting. They document that acquiring firms use stock splits to manipulate their equity values prior to the acquisition announcements, especially when the method of payment is in stock and when the deals are large. Due to the positive price reaction after split announcement acquirer shareholders gain wealth at the expense of target shareholders. Another possibility is that, managers of
4 5
Weld et al. (2009) provides an excellent review of the literature on these (and other) explanations for stock splits. See Fama et al. (1969).
Page 14 of 16
overvalued firms split to serve their own interest, potentially at the expense of their shareholders. Devos et al., (2015) find that CEOs manipulate the timing of the stock splits in order to increase CEO compensation through executive stock and option grants. Typically stock options are granted with an exercise price equal to the closing price of the grant announcement day. Using a sample of stock splits where the CEOs stock option grants occur within twenty one days surrounding split announcement, they find that majority (80%) of these grants occur on or before the split announcement day. They also find significantly higher levels of selling (purchasing) activity immediately after (before) split announcement. Examining the role of overvaluation in firms split decision is beyond the scope of this paper. Future research may address this area.
5. Conclusion This study investigates the proposition of signaling hypothesis which states that stock split announcements convey favorable inside information and therefore signal undervaluation of the splitting firms. Using three different mispricing measures, our results indicate otherwise. Contradictory to the signaling hypothesis, our findings show that split firms are overvalued in pre-announcement years and that the overvaluation reaches its peak in the split announcement year as compared to the years surrounding the split announcements. Moreover, overvaluation gradually declines in post-split announcement years across all three methods. Our results support the idea that stock splits do not signal undervaluation to the market.
Acknowledgments We thank Erik Devos and Rishav Bista for helpful comments.
Page 15 of 16
References Asquith, P., Healy, P. and Palepu, K. (1989), “Earnings and Stock Splits”, The Accounting Review, Vol. 64 No. 3, pp. 387-403. Brennan, M. and Copeland, T. (1988b), “Stock splits, stock prices and transaction costs”, Journal of Financial Economics, Vol. 22, No. 1, pp. 83-101. Desai, H. and Jain, P. (1997), “Long-Run Common Stock Returns Following Stock Splits and Reverse Splits”, Journal of Business, Vol. 70, No. 3, pp. 409-433. Devos, E., Elliott, W. B. and Warr, R. S. (2015), “CEO opportunism?: Option grants and stock trades around stock splits”, Journal of Accounting and Economics, Vol. 60 No. 1, pp. 18-35. D’Mello, R., and Shroff, P. (2000) ‘Equity valuation and decisions related to repurchase tender offers: an empirical investigation’, The Journal of Finance, Vol. 55 No. 5, pp. 2399-2424. Dyl, E. and Elliott, W. (2006), “The share price puzzle”, The Journal of Business, Vol. 79 No. 4, pp. 2045-2066. Fama, E., Fisher, L., Jensen, M. and Roll, R. (1969), “The adjustment of stock prices to new information”, International Economic Review, Vol. 10 No. 1, pp. 1-21. Grinblatt, M., Masulis, R. and Titman, S. (1984), “The Valuation Effects of Stock Splits and Stock Dividends”, Journal of Financial Economics, Vol. 13 No. 4, pp.461-490. Guo, S., Liu, M. H. and Song, W. (2008), “Stock Splits as a Manipulation tool: Evidence from Mergers and Acquisitions”, Financial Management, Vol. 37 No. 4, pp. 695-712. Huang, G., Liano, K., Manakyan, H. and Pan, M. (2008), “The Information Content of Multiple Stock Splits”, Financial Review, Vol. 3, No. 4, pp. 543-567. Ikenberry, D., Rankine, G. and Stice, E. (1996), “What do stock splits really signal?”, Journal of Financial and Quantitative Analysis, Vol. 31, No. 3, pp. 357-375. Ikenberry, D. and Ramnath, S. (2002), “Underreaction to Self-Selected News Events: The Case of Stock Splits”, Review of Financial Studies, Vol. 15 No. 2, pp. 489-526. Lakonishok, J. and Lev B. (1987), “Stock splits and stock dividends: Why, who, and when”, The Journal of Finance, Vol. 42 No. 4, pp. 913-932. Lin, J., Singh, A. and Yu, W. (2009), “Stock splits, trading continuity, and the cost of equity capital”, Journal of Financial Economics, Vol. 93, No. 3, pp. 474-489. McNichols, M. and Dravid, A. (1990), “Stock dividends, stock splits, and signaling”, The Journal of Finance, Vol. 45 No. 3, pp. 857-879. Rhodes-Kropf, M., Robinson, D. and Viswanathan, S. (2005), “Valuation waves and merger activity: The empirical evidence”, Journal of Financial Economics, Vol. 77 No. 3, pp. 561–603. Weld, W., Michaely, R., Thaler, R. and Benartzi, S. (2009), “The nominal share price puzzle”, Journal of Economic Perspectives, Vol. 23 No. 2, pp. 121-142.
Page 16 of 16